CONCLUSIONS 



461 



where dosage was maintained (41). These curves can be described as 

 showing mortahty rates constantly increasing with time. They fail to 

 follow a logarithmic trend as well as the data discussed before and are 

 plotted in linear fashion (Fig. 14). Single-dose data, not shown here, 

 also do not show any evidence of a subsiding tumor lethality rate up to 

 the life span of the experiments. This may be explained as a cumula- 

 tive acquisition of a tendency to tumor development not limited in time, 



100 200 300 



400 500 



Days 



600 700 800 900 



Fig. 14. Rate of morbidity from bone tumors in Carworth female mice injected with 



Sr*^. Retained do.se was maintained constant on the average throughout hfe by 



repeated injection. The points for each treatment group labelled by the size of the 



monthly injected dose in microcuries per gram. 



with a resultant accumulation of tumors as the square of time. Evans 

 has pointed out, however, that this may be the early part of a Gaussian 

 distribution with time dimensions greatly exceeding the period of ob- 

 servation (52) and the virtual disappearance of the human radium sar- 

 coma trend in the past few years indicates that in this case also an 

 actual limit in time (but on the order of several years) may exist. 



Conclusions 



The conclusions to this analysis will be stated in terms of a series of 

 suggestions as to where it appears that maximum effort is desirable in 

 order to answer the most important questions. 



1. We should examine most carefully the characteristic parts of the 

 lethality function (Fig. 7) : (a) the plateaus, where the more marked dif- 



